ckulla / xdoc Goto Github PK

I would like to be able to ignore a missing installation of pdfLatex. Otherwise the entire build of a project will fail due to a missing pdfLatex installation. I work with Linux and Windows. On Linux pdfLatex is installed, on windows it is not. I still want the windows build to succeed.

The model and xtext grammar already have elements for a glossary, but imho it is not fully implemented. Needs some investigation.

The generators should use composite objects for certain tasks (configured by guice). One can then for example provide a company style title page, header and footer.

For latex output, the width of images (i.e. figures) is computed by the helper class PDFImageUtil. In my environment, this always ends up with a width of 0.7 fraction of textwidth. How can this be influenced? I guess it would be useful to provide an optional DSL extension to xDoc to specify a width proposal for a figure (e.g., as a fraction of the textwidth).

Create a documentation of Xdoc based on Xdoc.

The pdfLatex executor should check whether all required tex packages are installed. If one is missing it should be installed automatically.

The title page of documentation generated from xDoc via Latex looks very "puristic". A nice option would be to be able to add some image, e.g. a logo, a screenshot or some other picture (maybe a photo of the author :-) ). This is very flexible and allows some individualization.

In latex, this could be converted to:

\usepackage{titlepic} <== as part of the usepackage list

...

\titlepic{\includegraphics{images/niceTitleImage.png}} <== after the \author... line

Right now the project structure is somewhat confusing, as there is a generator project, but the generator sare actually in the main project. My understanding is that the generator project provides old versions of the generators (written in xpand). Are those sill required? I would like to remove them to have a cleaner project structure.

The tests are currently in a bad condition. This should be fixed so new features can be tested easily.

Atm the project "org.eclipse.xtext.xdoc.updatesite" is contained in "org.eclipse.xtext.xdoc.releng". This confuses users.

With all generators I mean the html, eclipse help and latex generator.

The project should have a continuous build with travis (easy to setup).

With the xDoc=>html conversion, table are layouted by the html engine, which looks quite nice. For xDoc=>latex, a table is layouted by the xDoc=>latex converter with equally sized columns, which often looks ugly (esp. if some columns contain much less data than others). This should be improved.

How? Maybe there should be a heuristics "narrow column detector", which detects small columns and give them a width which is 50% of a "normal" column. Or the xdoc DSL could be extended to give some hints by the user. Don't know which option is better.